|Publication number||US7761480 B2|
|Application number||US 10/565,400|
|Publication date||Jul 20, 2010|
|Filing date||Jul 22, 2004|
|Priority date||Jul 22, 2003|
|Also published as||CA2533167A1, CN1856790A, EP1654679A2, EP1654679A4, US20060248045, US20100287158, WO2005008358A2, WO2005008358A3|
|Publication number||10565400, 565400, PCT/2004/667, PCT/IL/2004/000667, PCT/IL/2004/00667, PCT/IL/4/000667, PCT/IL/4/00667, PCT/IL2004/000667, PCT/IL2004/00667, PCT/IL2004000667, PCT/IL200400667, PCT/IL4/000667, PCT/IL4/00667, PCT/IL4000667, PCT/IL400667, US 7761480 B2, US 7761480B2, US-B2-7761480, US7761480 B2, US7761480B2|
|Inventors||Zev Toledano, Jair Jehuda|
|Original Assignee||Kinor Technologies Inc.|
|Export Citation||BiBTeX, EndNote, RefMan|
|Patent Citations (20), Non-Patent Citations (5), Referenced by (43), Classifications (14), Legal Events (3)|
|External Links: USPTO, USPTO Assignment, Espacenet|
The above-referenced application is the U.S. National Phase of International Application No. PCT/IL2004/000667, filed Jul. 22, 2004, which claims priority from U.S. Provisional Patent Application No. 60/489,768 filed Jul. 22, 2003. The International application was published on Jan. 27, 2005 as WO 2005/008358 A3 under PCT article 21(2).
The present invention relates generally to methods and systems for providing access to information, and specifically to federated networks for providing integrated access to multiple, diverse information sources.
There is a recognized need for data access systems and analytical tools that provide semantic interoperability among different information sources in a given application domain, both within a particular enterprise and across enterprise boundaries. Some tools that have been developed for this purpose make use of ontologies. An ontology is a structured vocabulary that represents the schematic metadata of a particular application domain. The ontology provides a unified, semantic model of the information in the domain, including both the types of entities that the information may include and relationships among the entities. The ontology allows users to express query concepts and relationships in high-level terms, which are then translated by appropriate agents into lower-level database schemata and semantic analyses.
One of the early tools of this sort was InfoSleuth™, developed at MCC (Microelectronics and Computer Technology Corporation, Austin, Tex.). InfoSleuth is described, for example, in an article by Fowler et al., entitled “Agent-Based Semantic Interoperability in InfoSleuth,” SIGMOD Record 28:1 (March, 1999), pp. 60-67, which is incorporated herein by reference. InfoSleuth is an agent-based system, in which a set, or community, of agents collaborate at a semantic level to execute information gathering and analysis tasks. The underlying information sources can be diverse in both structure and content. The agents, which are coded in Java™, communicate at the semantic level over ontologies using a Knowledge Query Manipulation Language (KQML). Agent types defined by InfoSleuth include:
Agents communicate and determine each other's capabilities using a shared ontological model of information management. The ontology provides the semantic framework for information activities in the domain of interest to the user. Semantic brokering allows agents to advertise their capabilities and to identify potential collaborators based on their advertised capabilities. The user may access the resources of the agent community from any location, and need know nothing about the physical location or structural characteristics of any resource.
Another method for distributed query handling is described by Wynblatt et al., in U.S. Patent Application Publication US 2002/0143755, whose disclosure is incorporated herein by reference. According to this method, a traditional database query is converted into network messages, which are routed to those data sources that have relevant data. The messages may be routed either directly or through designated query nodes. The data sources then send reply messages either directly to the originator of the query or via designated join nodes. In some embodiments, the data sources may be able to perform local join operations. The system collects the reply messages, and the messages that meet the requirements of the query are sent back to the query originator for presentation as a traditional database result.
Unicorn Solutions Inc. (New York, N.Y.) offers a Semantic Information Management (SIM) System, which it describes as a comprehensive platform for managing and integrating enterprise information resources. The system combines metadata repository, information modeling, hub-and-spoke mapping, and automated data transformation script generation capabilities. It is said to provide customers with a seamless business view by relating disparate data formats and interfaces to an information model that describes the business, its component parts, and all relationships. The system is described further in an article by Schreiber, entitled “Semantic Information Management (SIM): Solving the Enterprise Data Problem by Managing Data Based on its Business Meaning” (2003), which is available at www.unicorn.com and is incorporated herein by reference.
Various aspects of the Unicorn system are described in the patent literature. For example, U.S. Patent Application Publication US 2003/0163597, to Hellman et al., describes a method and system for collaborative ontology modeling, for use in building up an ontology from individual ontology efforts distributed over the Web. U.S. Patent Application Publication US 2004/0093344, to Berger et al., describes a method for mapping data schemas into an ontology model. U.S. Patent Application Publication US 2003/0101170, to Edelstein et al., describes a data query system using a central ontology model, in which a query processor generates a query in a data schema query language corresponding to a specified query expressed in an ontology query language. U.S. Patent Application Publication US 2003/0163450, to Borenstein et al., describes a method for providing a semantic registry for Web services and other services, based on an ontology model. The method is said to enable dynamic Web service integration by overcoming problems of semantic inconsistency. The disclosures of all the above patent application publications are incorporated herein by reference.
Embodiments of the present invention provide tools for use in a distributed, federated system for data management and access, based on a domain-specific ontology. In the embodiments disclosed hereinbelow, the system comprises a grid of agents and gateways, which communicate with one another over a communication network by exchanging semantic, ontology-based messages. The agents share a common, distributed platform, as well as a common directory of the ontology, data sources and rules that apply in the federated system. The gateways comprise wrappers, for interfacing between data sources and the ontology, as well as portals, through which users may address queries to the system. Upon receiving a user query, the portal directs the query to one of the agents, which develops a query plan based on the availability of necessary data and system resources for responding to the query. After collecting the required information in accordance with the query plan, the agent returns the query response to the user portal.
In embodiments of the present invention, the ontology accommodates logical rules that are applicable to the entities in the ontology, as well as to the attributes of and relationships between the entities. In some embodiments, the agent handling a query uses these rules in generating the query plan. For example, the agent may use the rules to determine which set of data fields, in which data sources, are most likely to give complete and correct data in response to a given query, and possibly also to validate the data that are collected in executing the query plan. The agent may also use the rules to determine the key according to which records from different sources should be joined, and whether the join should be performed by the agent or should be “pushed down” to the gateway for execution in order to optimize response or conserve system resources.
In some embodiments of the present invention, the ontological rules comprise access rules, indicating which user roles are authorized to read information in a given data source or set of sources. The access rules may pertain to particular data sources, to particular entities in the sources, or to particular attributes of a given type of entities. The access rules are expressed in unified terms, which may be maintained by a community of owners in the form of a community ontology that includes classes of roles, sources, clients, and authentication methods. These features of the ontology are thus useful in providing a unified model under which different data owners in a federated system may determine the access rules that are applicable to their data, such as which clients are trusted under which circumstances. The use of abstracted classes in the model, such as in a government database source, also enables clients and users to determine the scope of sources to be used without knowing each source specifically.
Typically, the access rules for each source are published together with additional source metadata in a grid-wide directory. Hence, the query agents can read the access rules immediately, without the need for central management of access authorizations. Since the access rules are already contained in the directory, there is no need for the agents to request authorization for each resource that is to be accessed (unless specifically required by the source owner). Furthermore, the agents may use the access rules in developing query plans so that the plan includes only those sources of information that the requesting user is authorized to access.
In some embodiments, access to each data source is logged along with the credentials of the user requesting the data. Thus, access policies are not only automatically enforced, but they are also recorded in a manner that supports automated auditing. Since the same mechanisms serve all grid-based client services, the services provided by the grid are automatically “regulated.”
In some embodiments of the present invention, a network utility gathers and updates information regarding the topology and status of resources in the system. Based on this information, the agents and gateways are able to discover and integrate new sources of information as they become available, as well as to determine the level of computing and communication loads on different agents and data sources. Thus, in making and executing the query plans, agents may use the topology and status information in order to optimize response time and dynamically load-balance the resources of the system.
In some embodiments, the grid functions as an application-level virtual private network, which links and provides services to the member sites and organizations. This virtual private network is internally managed by distributed functions of the grid, so that individual member sites need not explicitly provision for or coordinate their own virtual private networks with the other member sites. Thus, agents in the grid may even balance the overall load among resources that belong to different organizations within the federated system. For this purpose, the grid directory may comprise entities and attributes relating to performance of application-related tasks, which are used by the agents in monitoring resource performance and applying resource-use policies. Grid robustness and agility are enhanced by the use of dynamic, grid-generated query plans, as described above, and by the ability to balance loads across the grid. Altered grid topology, sources, and usage policies are automatically accommodated by generating new plans and rebalancing loads. Other grid conditions, such as fallen nodes, inaccessible sources, and revoked certificates, may be handled in similar fashion.
Furthermore, the fact that elements of the grid (agents and gateways) communicate with one another using a set of known, semantic, application-level messages may be used in securing communications among these elements against intrusions and malicious traffic. Thus, in some embodiments of the present invention, agents and/or gateways apply a semantic filter to each packet that they receive over the network and accept only packets that are compatible with the system ontology. Packets that do not meet the expected semantic criteria are discarded. Therefore, even if one of the elements of the grid is attacked by malicious traffic, the element will not permit the traffic to propagate into the grid. Moreover, gateways can examine each data element in the semantic payload of each packet to determine whether to pass it on or nullify it.
Although embodiments of the present invention are described herein with reference to information access and query handling, the principles of the present invention may similarly be applied to provision of other types of resources, such as Web services, over a network based on a semantic model.
There is therefore provided, in accordance with an embodiment of the present invention, a method for data access, including:
defining an ontology for application to a set of diverse data sources including data having predefined semantics;
associating with the ontology one or more logical rules applicable to the semantics of the data in the data sources;
receiving a query from a user regarding the data;
determining a query plan for responding to the query by selecting one or more of the data sources responsively to the ontology and by identifying an operation to be applied to the data responsively to the applicable logical rules; and
generating a response to the query in accordance with the query plan.
In a disclosed embodiment, the logical rules include a validation rule, and the query plan includes validating the data from at least one of the data sources responsively to the validation rule.
Additionally or alternatively, the logical rules include a mapping rule, such that data sources that match these rules can be automatically mapped to the ontology when the sources are primed and attached to the grid. Further additionally or alternatively, the logical rules include a joining rule, and the query plan includes selecting a key responsively to the joining rule, and joining the data from two or more of the data sources using the key. Typically, selecting the key includes analyzing the data so as to select one or more fields in the two or more of the data sources for use as the key so as to provide a desired statistical probability that the data will be joined correctly.
Yet further additionally or alternatively, the logical rules include a transformation rule, and the query plan includes transforming the data in at least one of the data sources from a first value that is held in the at least one of the data sources to a second value responsively to the transformation rule.
Still further additionally or alternatively, the logical rules include a business logic rule, and the query plan includes processing the data from at least one of the sources responsively to the business logic rule.
The logic rules may also include an access rule, and the query plan may include selecting at least one of the data sources for use in generating the response responsively to the access rule as applied to the user and client who submitted the query.
In some embodiments, defining the ontology includes associating a respective wrapper with each of the data sources, so as to transform the data from each of the data sources from a native format to an ontological format determined by the ontology, and generating the response includes applying the operation using the wrapper, and then reporting the data from the wrapper to a hub that links the data sources following application of the operation.
In one embodiment, the operation applied by the wrapper includes joining the data from two or more of the data sources.
Additionally or alternatively, the operation applied by the wrapper includes mapping values of the data. For example, mapping the values may include normalizing the data from a native representation to an ontological representation.
Further additionally or alternatively, the query plan includes a group of sub-queries, and generating the response includes sending the sub-queries from an agent running on the hub to respective wrappers of a plurality of the data sources, and combining the data reported from the wrappers in order to produce the response. Typically, sending the sub-queries includes invoking two or more of the wrappers to operate in parallel.
In a disclosed embodiment, associating the respective wrapper includes distributing an advertisement of each of the data sources in accordance with the ontology, and determining the query plan includes discovering each of the data sources responsively to the advertisement, and building the query plan based on the discovered data sources.
In an aspect of the invention, reporting the data includes sending data packets over a network, the packets including semantic content in a form determined by the ontology, and upon receipt of the data packets at the hub, verifying legitimacy of the packets responsively to the semantic content and nullifying semantic elements that must be filtered.
Typically, reporting the data includes streaming the data from the wrapper to a specified storage location. Alternatively, reporting the data includes moving the data in a block operation from the wrapper to a specified storage location.
In another aspect of the invention, determining the query plan includes collecting information regarding a topology and performance characteristics of the data sources, and selecting, responsively to the information, the data sources to be used in responding to the query.
There is also provided, in accordance with an embodiment of the present invention, a method for providing a user with access to a set of diverse information resources, which are configured to provide information with predefined semantics, the method including:
defining an ontology for application to the information provided by the set of diverse information resources;
associating with the ontology one or more logical rules applicable to the semantics of the information;
receiving a request from the user to access the information;
determining a plan for responding to the request by selecting one or more of the information resources responsively to the ontology and by identifying an operation to be applied to the information responsively to the applicable logical rules; and
generating a response to the request in accordance with the plan.
In a disclosed embodiment, the request includes a query for data held by the information resources.
In another embodiment, the request includes a subscription request, and generating the response includes providing a succession of responses over a period of time responsively to the subscription request.
In yet another embodiment, the information resources are configured to provide Web services, and the request specifies one or more of the Web services to be provided to the user.
In still another embodiment, the request specifies data to be written to a data repository associated with one or more of the information resources.
In a further embodiment, the request specifies a transaction to be carried out and recorded by one or more of the information resources.
In an additional embodiment, the request specifies a process to be carried out by one or more of the information resources. Typically, the request specifies an event, and generating the response includes carrying out the specified process responsively to an occurrence of the event.
There is additionally provided, in accordance with an embodiment of the present invention, a method for data access, including:
defining an ontology for application to a set of diverse data sources including data;
defining data access rights with respect to the ontology; and
controlling user access to the data responsively to the ontology of the data and the access rights applicable thereto.
In a disclosed embodiment, defining the ontology includes specifying a user ontology, and defining the data access rights includes assigning a classification to a user according to the user ontology, and controlling the user access includes comparing the classification to the access rights applicable to the data. Typically, the diverse data sources are distributed among a set of autonomous organizations including at least first and second organizations, and assigning the classification includes classifying the user according to an organizational affiliation of the user so as to control access by users in the first organization to the data sources held by the second organization.
Typically, controlling the user access includes receiving a query from a user to access the data in the data sources, determining a query plan for responding to the query by selecting one or more of the data sources responsively to the ontology such that the access rights permit the user to access the data in the one or more of the data sources, and generating a response to the query in accordance with the query plan.
There is further provided, in accordance with an embodiment of the present invention, a method for data access, including:
defining a set of data resources providing access to data;
collecting information regarding a topology and performance characteristics of the data resources;
receiving a query from a user regarding the data;
determining a query plan responsively to the query and to the information regarding the topology and performance characteristics; and
generating a response to the query in accordance with the query plan.
Typically, collecting the information includes measuring respective load levels of two or more of the data resources, and determining the query plan includes selecting one of the data resources so as to balance the load levels.
In a disclosed embodiment, the data resources are distributed among a set of autonomous organizations including at least first and second organizations, wherein the user submitting the query belongs to the first organization, and wherein determining the query plan includes selecting, responsively to the performance characteristics, one of the data resources of the second organization for use in responding to the query.
There is moreover provided, in accordance with an embodiment of the present invention, a method for exchange of information, including:
establishing a virtual private network among a plurality of nodes, including at least first and second nodes, which are configured to communicate with one another over an underlying public physical network;
defining a semantic communication model for conveying data packets among the nodes in the virtual private network, responsively to an ontology of the information;
sending a data packet over the virtual private network from the first node to the second node; and
filtering the data packet against the semantic communication model at the second node, so as to verify that the data packet is legitimate and nullify semantic elements that must be filtered.
Typically, defining the semantic communication model includes defining a closed set of semantic messages that may be carried by data packets in the virtual private network.
In one embodiment, the nodes are distributed among a set of autonomous organizations.
In a disclosed embodiment, the nodes include gateway nodes, which are configured to communicate with clients and data sources using native data formats, and to translate the native data formats to the semantic communication model for communication over the virtual private network. Typically, the nodes further includes hub nodes, and establishing the virtual private network includes configuring the gateway nodes to communicate with the hub nodes using the semantic communication model.
There is furthermore provided, in accordance with an embodiment of the present invention, apparatus for data access, including a hub processor, which is adapted to receive a definition of an ontology for application to a set of diverse data sources including data having predefined semantics, and to associate with the ontology one or more logical rules applicable to the semantics of the data in the data sources, and which is further adapted, upon receiving a query from a user regarding the data, to determine a query plan for responding to the query by selecting one or more of the data sources responsively to the ontology and by identifying an operation to be applied to the data responsively to the applicable logical rules, and to generate a response to the query in accordance with the query plan.
There is also provided, in accordance with an embodiment of the present invention, apparatus for providing a user with access to a set of diverse information resources, which are configured to provide information with predefined semantics, the apparatus including a hub processor, which is adapted to receive a definition of an ontology for application to the information provided by the set of diverse information resources and to associate with the ontology one or more logical rules applicable to the semantics of the information, and which is further adapted, upon receiving a request from the user to access the information, to determine a plan for responding to the request by selecting one or more of the information resources responsively to the ontology and by identifying an operation to be applied to the information responsively to the applicable logical rules, and to generate a response to the request in accordance with the plan.
There is additionally provided, in accordance with an embodiment of the present invention, apparatus for data access, including a hub processor, which is adapted to receive a definition of an ontology for application to a set of diverse data sources including data and a definition of data access rights with respect to the ontology, and which is adapted to control user access to the data responsively to the ontology of the data and the access rights applicable thereto.
There is moreover provided, in accordance with an embodiment of the present invention, apparatus for data access, including a hub processor, which is adapted to receive a definition of a set of data resources providing access to data, and to collect information regarding a topology and performance characteristics of the data resources, and which is further adapted, upon receiving a query from a user regarding the data, to determine a query plan responsively to the query and to the information regarding the topology and performance characteristics, and to generate a response to the query in accordance with the query plan.
There is furthermore provided, in accordance with an embodiment of the present invention, apparatus for exchange of information, including a plurality of computing nodes, which include at least first and second nodes, and which are linked to communicate over a virtual private network running over an underlying public physical network, and which are configured to exchange data packets over the virtual private network in accordance with a semantic communication model, which is defined responsively to an ontology of the information, wherein at least the second node is adapted, upon receiving a data packet over the virtual private network from the first node, to filter the data packet against the semantic communication model so as to verify that the data packet is legitimate.
There is further provided, in accordance with an embodiment of the present invention, a computer software product, including a computer-readable medium in which program instructions are stored, which instructions, when read by a computer, cause the computer to receive a definition of an ontology for application to a set of diverse data sources including data having predefined semantics, and to associate with the ontology one or more logical rules applicable to the semantics of the data in the data sources, and further cause the computer, upon receiving a query from a user regarding the data, to determine a query plan for responding to the query by selecting one or more of the data sources responsively to the ontology and by identifying an operation to be applied to the data responsively to the applicable logical rules, and to generate a response to the query in accordance with the query plan.
There is also provided, in accordance with an embodiment of the present invention, a computer software product, for providing a user with access to a set of diverse information resources, which are configured to provide information with predefined semantics, the product including a computer-readable medium in which program instructions are stored, which instructions, when read by a computer, cause the computer to receive a definition of an ontology for application to the information provided by the set of diverse information resources and to associate with the ontology one or more logical rules applicable to the semantics of the information, and further cause the computer, upon receiving a request from the user to access the information, to determine a plan for responding to the request by selecting one or more of the information resources responsively to the ontology and by identifying an operation to be applied to the information responsively to the applicable logical rules, and to generate a response to the request in accordance with the plan.
There is additionally provided, in accordance with an embodiment of the present invention, a computer software product, including a computer-readable medium in which program instructions are stored, which instructions, when read by a computer, cause the computer to receive a definition of an ontology for application to a set of diverse data sources including data and a definition of data access rights with respect to the ontology, and to control user access to the data responsively to the ontology of the data and the access rights applicable thereto.
There is further provided, in accordance with an embodiment of the present invention, a computer software product, including a computer-readable medium in which program instructions are stored, which instructions, when read by a computer, cause the computer to receive a definition of a set of data resources providing access to data, and to collect information regarding a topology and performance characteristics of the data resources, and further cause the computer, upon receiving a query from a user regarding the data, to determine a query plan responsively to the query and to the information regarding the topology and performance characteristics, and to generate a response to the query in accordance with the query plan.
There is moreover provided, in accordance with an embodiment of the present invention, a computer software product, including a computer-readable medium in which program instructions are stored, which instructions, when read by a group of computing nodes that includes at least first and second nodes, linked to communicate over a physical public network, cause the computing nodes to communicate in a virtual private network by exchanging data packets over the public physical network in accordance with a semantic communication model, which is defined responsively to an ontology of the information, wherein the instructions cause at least the second node, upon receiving a data packet over the virtual private network from the first node, to filter the data packet against the semantic communication model so as to verify that the data packet is legitimate.
The present invention will be more fully understood from the following detailed description of the embodiments thereof, taken together with the drawings in which:
On the other hand, the information to be shared within system 20 is typically held in many different data formats, at different locations within different organizations, with their own information models, structures and representations. To further complicate the situation, different databases frequently use different names to denote the same information. For example, hospital 26 may lists the names of persons using its services in a database field labeled “patients,” while HMO 24 refers to the same names as “members.” In addition, much of the information held by the organizations in system 20 is of a confidential nature. Therefore, in sharing information with other parties, each organization must ensure that access to information by other parties in the system does not violate its own confidentiality policies. Although these problems are illustrated here with respect to a healthcare-related system, it will be appreciated that similar problems arise in collaboration between organizations of other sorts, such as business enterprises or government agencies.
Embodiments of the present invention addresses the needs and problems outlined above by linking the participating organizations to exchange information over an internal, grid-managed virtual private network 42, with a semantic, ontology-based foundation. Physically, network 42 may run over existing, generic wide-area networks (WANs) 36, such as the Internet, and local area networks (LANs) 38. Logically, however, the users (clients) and data sources within system 20 are connected through secure, semantic links of virtual network 42 to a backbone of hubs 40. Although these hub nodes are shown in
As described in detail hereinbelow, embodiments of the present invention use several key inventive elements in addressing the needs of system 20:
(a) federated communities managing their ontologies;
(b) federated owners (“sites”) managing their clients and sources;
(c) a fluid agent-based architecture that supports a growing set of semantic client services in a scalable and robust manner; and
(d) a distributed grid platform of gateways and hubs (referred to hereinbelow simply as the “grid”) that hosts the agents and ontologies and spans any number of sites for any number of communities in a secure, self-managed and self-adapting manner (thus creating, in effect a “grid of grids”).
In describing embodiments the present invention, it is helpful to distinguish between priming, background and foreground tasks:
(a) Priming tasks are carried out by community and site administrators when assembling, growing, and evolving the grid. Examples of such tasks include wrapping and attaching a new source, adding a new site, altering a usage policy.
(b) Foreground tasks are carried out by the grid to respond to client requests during run-time, for example, executing a client query or subscription.
(c) Background tasks are carried out by the grid to maintain, self-adapt, and improve services. Examples of background tasks include periodically checking source integrity, producing auxiliary tables for more efficient joins, dynamically indexing internal tables, and reconfiguring agents to better address dynamic global requirements.
Sites participating in system 20 are linked to network 42 by respective gateways 50. Gateways have two possible functions:
Query agents 60 run on hubs 40, wherein each hub may run one or more of these agents. Agents 60 receive queries from portals 52 and “compile” the queries into query plans that identify potential sources of the data. The query agents then carry out the plans by sending appropriate messages, referred to hereinbelow as sub-queries, to wrappers 56 requesting that they supply the required data from respective sources 58. When execution of the query plan is completed, agent 60 reports the results back to the requesting portal 52. The query planning and execution process is described in greater detail hereinbelow. Hubs 40 may also run agents of other types, such as subscription agents, i.e., agents that listen continuously for source, data, or grid-related events of a specified type in system 20, and then report the results of these events to a client. Further agent types are enumerated below.
In planning and executing queries, the query agent on hub 40 is assisted and supported by other hub utilities (not shown in the figure), which may also be implemented as agents. These utilities include the following:
Gateways 50 may serve any number of communities, but each hub 40 typically serves a specific community. Each community defines the nature of client services to be supported, possibly differentiating between different service priorities. Each service and priority is handled by an appropriate set of service agents 60, 64, for example, query and subscription agents, which communicate via messaging services 66 with portal and wrapper agents as needed. Multiple agents at different nodes can share in the handling of any specific service and priority by subscribing to a shared message queue 68 of pending requests. Similarly, portal agents 65 submitting multiple types of client service requests publish the requests to message queues 68 designated for handling each type of service request. Message queues are dynamically maintained by the distributed grid platform to facilitate changes in the current configuration of services and agents.
In this embodiment, each gateway 50 is configured to connect to a specific set of hubs 40. Each community maintains a list of hubs serving that community so that the hubs can readily discover each other. A dynamic grid-wide directory 67 maintains the current state of nodes (gateways and hubs) and connectivity, as well as a registry of all sources, clients, and agents accessible via these nodes. Agents and other utilities ensure that the dynamic directory is kept up to date as part of their background activities. As noted earlier, all grid nodes and components are mutually authenticated, and an internal virtual private network is maintained to produce a secure grid substrate.
When one of clients 54 submits a service request to one of portal agents 65 on its gateway 50, this request is queued to all hub agents 60, 64 that have subscribed to such requests. An agent picks up the request, produces a concrete plan for executing it, and then executes it with the assistance of other agents, such as the source wrapper agents 65 needed to produce the results. In such a case, each wrapper agent receives an appropriate sub-request via its queue. The results of the sub-request are stored directly in storage 63 at a predetermined location. The servicing hub agent then collects these partial results to produce final results in the same storage location or any other storage location. A link to the final results is then sent to the originating portal agent via its message queue.
As implied by this description, the grid topology and consistency is fluid enough to accommodate change without disruption. Thus, for example, if a source becomes unavailable, its wrapper agent will update the directory. Thereafter, all agents with ongoing tasks that use this source can either produce a new plan of action without the source or return results with an appropriate exception (assuming the source plays a non-critical part in the current query plan). If a grid node fails, agents running on that node can be restarted at other nodes. A grid-wide replication service supports replicated components, including replicated directories and messages, so that there is no single point of failure.
Ontology 70 typically contains a hierarchy of classes, which provide a vocabulary for describing entities, such as people and vehicles, and their attributes, such as age and weight. (Ontology classes are commonly referred to as “entities” or “frames,” and their attributes are referred to as “slots.”) In the example shown in
As in object-oriented programming, inheritance may be used in ontology 70 to define multiple abstraction layers. Thus, trader 76 inherits the ontology of person 80, with sub-entities of name, birth date, social security number (SSN), etc. Each individual instance of a given entity, such as person 80, is a specific person or other object, with its own (presumably unique) set of attribute values.
Ontology 70 also defines relationships, by associating semantic labels with relationships that may exist between entities. For example, trader 76 makes trade 74, while trade 74 is traded by trader 76. Properties in the ontology define different aspects of the ontological slots and relationships, such as the slot types and cardinality, value ranges and statistics, and how one relationship is the inverse of another. Further general aspects of ontologies, including how ontologies may be created and their use in data management, are described in the references cited in the Background of the Invention. To permit data in sources 58 to be accessed via the ontological model of network 42, wrappers 56 map the metadata of the respective sources to corresponding entities and attributes of the ontology. Automated methods for carrying out this mapping are describe hereinbelow.
In embodiments of the present invention, the ontologies used in system 20 incorporate logical rules, which enable the ontology to capture generalized domain knowledge that can then be applied in query planning and execution. In an object-oriented framework, the rules may be expressed as methods associated with the object classes of the different entities in the ontology. Exemplary types of rules may include the following:
Agent 60 relies on a unified metadata model 90 in compiling query plans. The metadata model includes an ontology 92, as described above, and a source directory 94. A semi-automated ontology builder 96 may be used to generate and modify ontology 92, typically under control of a system manager. The ontology builder then maps data structures in sources 58 to the terms in the ontology. A source modeler 98 maps sources 58 themselves to source directory 94, based on the ontology of sources in system 20. User access privileges are similarly maintained in the source directory in ontological terms.
Queries composed by clients 54 are submitted to a query server 100. The queries may be composed using a graphical user interface (GUI) 102, which enables users to browse and query all authorized data in ontological terms. In other words, the users are not limited to predetermined views of the data in sources 58, as in conventional database management systems, but may rather choose any combination of data and sources that they are authorized to access (based on the policies contained in the system ontology, as described above).
Query server 100 refers user queries to a query compiler 104, which composes a query plan for responding to each query. Query planning is described further hereinbelow with reference to
Once the query plan has been compiled, server 100 addresses the query messages to the appropriate sources 58, and thus builds the query response for client 54.
Portal 52 then selects a query agent 60 to carry out the query, at an agent selection step 112. The choice of query agent may take into account the resources available to the agent, its current work load, and its qualifications to deal with the given ontology of the query. Directory 67 (
Query compiler 104 parses and analyzes the query and uses the matching utility mentioned above to produce a subset of data sources 58 (or equivalently, a subset of wrappers 56) that are relevant to the query, at a data source identification step 114. Selection of the data sources depends on the ontological terms of the query, the requested scope of the data, the client and user credentials, and the source use policies. Specifically, for each candidate data source, the query compiler checks the client and/or user identity against the access policies that are included in the source ontology, as listed in ontology 92, at an access checking step 118. Any source for which the query compiler determines that the client or user does not have sufficient access rights is removed from the query plan.
The query compiler next examines the query and the matched subset of data sources in order to compile the ontological query into an optimized query plan that best exploits the available data sources, at a query compilation step 120. Further aspects of query compilation and optimization are described below with reference to a specific query example in
Query agent 60 then executes the plan, sending appropriate sub-query messages to wrappers 56 of sources selected in the plan, at a message distribution step 122. The messages typically comprise handles to the storage resources allocated for the responses. Wrapper invocations are initiated in parallel whenever possible, as illustrated in the figure.
Wrappers 56 use the source mappings to translate the ontological sub-queries into the appropriate native terms of local data sources 58. Details of this process are shown below in
The query plan may call for joining, filtering, sorting and additional actions to be carried out by the query agent upon subsets of results at various stages. The query agent keeps track of the wrapper notifications and uses DBMS capabilities of the storage utility to carry out these actions when the result subsets are ready. The query agent also handles exceptions, such as when results are not obtained within designated response times. The query plan may indicate which results are critical to the query and which are not, so that the query agent can respond accordingly. For example, when a sub-query requesting non-critical results fails, the results may simply be omitted from the query response. On the other hand, critical results, if unavailable from the original intended source, may be procured by agent 60 from another source, based on the information in metadata sources 90.
Query agent 60 checks periodically to determine whether the query plan has been completed, at a completion checking step 130. As long as further results are still required in order to complete the query plan, the query agent continues to collect and process data from wrappers 56, at a continued collection step 132. Upon completion of the query plan, the query agent caches the results in an appropriate storage location, at a caching step 134. The agent then provides portal 52 with a handle to the final results, at a handle passing step 136. The portal transforms the cached results into client terms as needed, and delivers the results in the form requested by the client. All the above actions are typically logged by the logging utility, along with the credentials of the user and client that submitted the query, for auditing and accountability purposes.
As noted previously, the sub-query from agent 60 is typically stated in the ontological terms that are used in network 42. Wrapper 56 maps the sub-query from the ontological “SQL” used by agent 60 to the native query format that is understood by the data source 58 that wrapper 56 serves, at a query mapping step 144. Although
If required by the query plan, wrapper 56 may locally perform a part of the data processing required by the plan, at a pushdown step 150. This step may be carried out using the processing capabilities of the local DBMS or using processing capabilities built into the wrapper. The step may include, for example, joining values from multiple tables held in data source 58 or, additionally or alternatively, performing arithmetic or string operations on data values. Some examples are shown in
After retrieving the data from source 58 and completing any required local processing, wrapper 56 normalizes the results in the appropriate ontological terms. The wrapper stores the results in the appropriate location using the above-mentioned storage service, at a data storage step 152 (corresponding to steps 124 and 126 in
Although the scenario above relates to handling SQL-type queries, in alternative embodiments of the present invention, system 20 is configured to support a comprehensive set of information services, such as data updates, subscriptions, events, and backups. Each service is typically carried out by an appropriate type of service agent. Requests from clients 54 for services of these types are submitted by portal 52 to the appropriate service agent, which collaborates with other agents and invokes utilities as needed.
The following are a few examples of alternative service types that may be supported by system 20:
The data sources available for responding to queries in this example include relational databases 170 and 172 and structured documents 174, 176, 178. It can be seen in the figure that the fields in the databases and documents have already been mapped by wrappers (not shown in this figure) to ontology 160. Thus, for example, a field that might have appeared in sales database 172 as the “purchase order number” has been mapped to the ID slot of sale entity 162, while another field labeled “catalog number” has been mapped to the ID slot of product entity 164.
A query 180 is now submitted to system 20, asking for a listing of products sold, by name and price, and the amounts and account numbers of sales in which these products have been sold. Query compiler 104 (
Selection of the keys to be used in joining data from different sources is a function of compiler 104. The compiler uses ontological rules in order to choose the optimal key to use in each case, so as to increase the likelihood of obtaining complete and unambiguous data.
For example, consider the case in which a query is submitted regarding a group of persons, requiring that query agent 60 provide contact data (such as address), personal data (such as age), and a vehicle registration number for each person. The data are assumed to be in different tables, and the query requires that they be joined for an authorized client. If all tables are contained in the same database, the source metadata will typically identify foreign keys for joining them. Otherwise, it is possible that the data are distributed among several sources, wherein each source has a field that has been mapped to the same distinct identifier, such as an employee number, or the unique sale ID number noted in the example above. In this case, the common identifying field or fields provide a solid base for joining the tables. Source directory 94 (
Otherwise, if not all the sources have fields mapped to distinct identifiers, compiler 104 may use rules contained in ontology 92 to find a key. For example, the ontology might reveal that a name and social security number can be transformed into a distinct identifier in this case, source fields mapped to the name and social security number can be used to join the relevant tables. Otherwise, compiler 104 may perform empirical tests to evaluate the statistical uniqueness of using the available non-distinct fields in these sources as the basis for joining. For example, the compiler may count the number of people in the target population that actually have the same name and age, and may then choose the most distinct combination as the key for joining the tables.
Even when distinct keys are available, erroneous data (such as misspelled names) and ambiguous formats (such as different U.S. and European date formats) can disrupt the joining process. Compiler 104 may use ontological rules and empirical testing to determine the likelihood that such cases exist. For example, the query agent may test for similar names that have identical social security numbers and home addresses. It may then construct an auxiliary join table to automatically translate similar names and other personal attributes into distinct person identifiers. Each entry in the table may also list the evaluated probability that the relationship it represents is valid. This auxiliary table can subsequently be used as the basis for joining data from the original tables, and also gives a well-defined measure of probability that the join was valid. Such join tables can be produced per class of entities or for specific subsets of each class.
Thus, for example, there may be several sources of person-related information, each with different column combinations that include different combinations or sub-combinations of name, social security number, address and telephone numbers. Using rules contained in the ontology, information in these tables could be correlated to identify which source records belong to which people. On this basis, a virtual person ID could be generated for each person, and an auxiliary table could be produced for each source containing the most probable person ID for each record and the confidence level of the person ID. Subsequent joins for person-related information could then use the grid-generated virtual IDs as long as the confidence levels exceed a threshold provided with the query.
New data sources may be added to system 20 on a “plug-and-play” basis. For this purpose, a wrapper is first created, in order to map the fields of the new source to the appropriate ontology, at a wrapper creation step 190. This step is typically carried out semi-automatically by a system manager, using a wrapper “wizard” program, as described below with reference to
Ontology builder 96 may also generate new ontology versions. For this purpose, the ontology builder applies a version management utility to automatically transform ontological expressions from one version to another. Similarly, source mappings to one ontology version are automatically upgraded in source directory 94 to new versions. Typically, ontological evolution consists mainly of expansion of the topology, with occasional renaming and merging of structures, so that forward and backward transformations between versions are relatively simple.
From time to time, the organizations participating in system 20 may choose to change use policies with respect to data sources 58 and other resources that they make available via network 42. Such policy changes are likewise advertised by wrappers 56, at a policy advertisement step 196, and are then discovered and recorded in source directory 94, at a policy discovery step 198.
During the normal course of activities of system 20, the above-mentioned logging utilities (which may run on both hubs 40 and gateways 50) track changes in the topology of network 42 and measure performance of various network functions, at a monitoring step 200. The monitoring semantics, like other functions of network 42, are based on a common, network-wide ontology, which defines entities such as queries submitted and answered, data transfer quantities and rates, duration of operations and timeouts. Activities are logged at the nodes at which they occur, so that the logs accumulate in a manner that does not disrupt messaging. Each log records each activity along with an identification of the request that generated it, so that a complete record of the handling of any particular request can be reconstructed end-to-end based on information held in the different logs. For the purpose of such logging, a network-wide system clock may be defined (despite the fact that the underlying physical network is asynchronous) by periodic distribution of appropriate timing messages, thus creating a “virtual time” that is uniform for all logs in system 20. As noted earlier, query agents 60 use the topology and performance data in order to perform load balancing among resources of system 20, even across boundaries of participating organizations. Ontology-based use policies ensure that the resources of one member organization will not be used by other organizations in the system beyond the level permitted by organization policies.
Similar models may be used for wrapping data resources of other types. For example, if the resource is a Web service, the source Web service description language (WSDL) model may be automatically loaded into the source modeler, in place of the SQL metadata. Similarly, if the resource is an XML database, an XQuery server may be used to identify available schemata for mapping by the source modeler. For unstructured text, analytical tools such as those offered by ClearForest (New York, N.Y.) can be used to add XML tags to the text, which can then be indexed like an XML database. Likewise, models of this type may be used for mapping other elements of system 20 to the appropriate ontologies.
Clients 220 and 250 may communicate with sources, such as source 224, only via the respective gateways and the backbone agents. All communications between the clients and the gateways are purely semantic in form, as are all communications between agents. An exemplary data packet 240 contains a message queue header 246 and a semantic payload 248. Header 246 comprises the logical addresses of the application-level source and destination of the packet (typically agents) and possibly other parameters. Payload 248 comprises a query, response or other system message. The contents of both header 246 and payload 248 are selected from a limited, predetermined set of possible semantic contents, which must conform to the specified ontology of system 20. This ontology is known to all of the agents.
Therefore, upon receiving a packet over network 42, the receiving agent can check the semantics of packet header 246 and payload 248 for conformance with the known ontology. In other words, the agent filters the header and payload against a limited set of known templates to which legitimate traffic must conform. Packets with non-conforming semantics are simply discarded, or filtered semantic elements are nullified. (By contrast, in generic IP networks, although a firewall or other intrusion protection device may check the packet payload for signatures of known malicious traffic types, the payload remains a “black box” of binary data, which cannot readily be diagnosed as benign or malicious.) Network 42 is thus substantially immune to malicious traffic attacks. Typically, for greater data security and intrusion prevention, header 246 and payload 248 are encrypted.
In a worst-case scenario, a malicious user, operating on client 250, for example, may “hack into” gateway 252, and may then bombard agent 230 with packets. Agent 230, however, will recognize that the packets do not have the proper semantics and will discard them. Even if the malicious user is successful in this manner in disrupting the performance of agent 230, the resultant traffic will not propagate further into network 42. Other clients, such as client 220, will still be able to access data sources through the remaining agents in the backbone grid.
Moreover, each client 220 and source 224 can determine which ontological elements it is willing to share and what security checks should be applied to each element before it is either sent or received. Since all shared information must correspond to ontological elements to flow through the grid, each owner is guaranteed that all incoming and outgoing flows comply with well-defined standards, regardless of the nature of the service being provided. This model is far less complex than the common alternative of screening binary information, which may assume any form for any number of clients and applications with no semantic basis.
As noted earlier, although the embodiments described above relate mainly to providing access to data sources, the principles of the present invention may similarly be applied in controlling access to and supply of other information resources, such as Web services, via a network. It will thus be appreciated that these embodiments are cited by way of example, and that the p resent invention is not limited to what has been particularly shown and described hereinabove. Rather, the scope of the present invention includes both combinations and subcombinations of the various features described hereinabove, as well as variations and modifications thereof which would occur to persons skilled in the art upon reading the foregoing description and which are not disclosed in the prior art.
|Cited Patent||Filing date||Publication date||Applicant||Title|
|US6442545||Jun 1, 1999||Aug 27, 2002||Clearforest Ltd.||Term-level text with mining with taxonomies|
|US6532469||Sep 20, 1999||Mar 11, 2003||Clearforest Corp.||Determining trends using text mining|
|US6604093||Dec 27, 1999||Aug 5, 2003||International Business Machines Corporation||Situation awareness system|
|US6636848||Jul 6, 2000||Oct 21, 2003||International Business Machines Corporation||Information search using knowledge agents|
|US6847974 *||Jul 25, 2001||Jan 25, 2005||Us Search.Com Inc||Method and apparatus for intelligent data assimilation|
|US7099885 *||May 25, 2001||Aug 29, 2006||Unicorn Solutions||Method and system for collaborative ontology modeling|
|US20010047485 *||Mar 5, 2001||Nov 29, 2001||Daniel Brown||Computer security system|
|US20020120618||Aug 20, 2001||Aug 29, 2002||Kazutomo Ushijima||Integrated database system and program storage medium|
|US20020138353||Mar 2, 2001||Sep 26, 2002||Zvi Schreiber||Method and system for analysis of database records having fields with sets|
|US20020143755||Mar 11, 2002||Oct 3, 2002||Siemens Technology-To-Business Center, Llc||System and methods for highly distributed wide-area data management of a network of data sources through a database interface|
|US20020194181||Jul 25, 2001||Dec 19, 2002||Wachtel David C.||Method and apparatus for intelligent data assimilation|
|US20030101170||May 31, 2002||May 29, 2003||Joseph Edelstein||Data query and location through a central ontology model|
|US20030120665||Mar 22, 2002||Jun 26, 2003||Joshua Fox||Run-time architecture for enterprise integration with transformation generation|
|US20030163450||Jan 9, 2003||Aug 28, 2003||Joram Borenstein||Brokering semantics between web services|
|US20030163597||May 25, 2001||Aug 28, 2003||Hellman Ziv Zalman||Method and system for collaborative ontology modeling|
|US20030177481||Nov 22, 2002||Sep 18, 2003||Amaru Ruth M.||Enterprise information unification|
|US20030179228||Jul 6, 2001||Sep 25, 2003||Schreiber Marcel Zvi||Instance browser for ontology|
|US20040093344||Aug 8, 2003||May 13, 2004||Ben Berger||Method and system for mapping enterprise data assets to a semantic information model|
|US20040093559||Aug 7, 2003||May 13, 2004||Ruth Amaru||Web client for viewing and interrogating enterprise data semantically|
|US20070027874 *||Sep 21, 2006||Feb 1, 2007||Kilian Stoffel||Computer-based method and apparatus for repurposing an ontology|
|1||Fowler et al. entitled "Agent-Based Semantic Interoperability in InfoSleuth" SIGMOD Record 28:1 (Mar. 1999), pp. 60-67.|
|3||Schreiber, "Semantic information management (SIM): Solving the enterprise data problem by managing data based on its business meaning", 2003 (available at www.unicorn.com).|
|4||Shah, A. "Source Specific Query Rewriting and Query Plan Generation for Merging XML-Based Semi-structured Dara in Mediation Systemes" University of Florida, May 2001, pp. 1-113, whole documents.|
|5||U.S. Appl. No. 60/489,768, filed Jul. 22, 2003.|
|Citing Patent||Filing date||Publication date||Applicant||Title|
|US8041570 *||May 31, 2005||Oct 18, 2011||Robert Bosch Corporation||Dialogue management using scripts|
|US8140556||Jan 20, 2009||Mar 20, 2012||Oracle International Corporation||Techniques for automated generation of queries for querying ontologies|
|US8214401||Feb 26, 2009||Jul 3, 2012||Oracle International Corporation||Techniques for automated generation of ontologies for enterprise applications|
|US8219572||Aug 29, 2008||Jul 10, 2012||Oracle International Corporation||System and method for searching enterprise application data|
|US8296317||Sep 15, 2008||Oct 23, 2012||Oracle International Corporation||Searchable object network|
|US8335778 *||Sep 17, 2008||Dec 18, 2012||Oracle International Corporation||System and method for semantic search in an enterprise application|
|US8375029 *||Oct 31, 2007||Feb 12, 2013||British Telecommunications Public Limited Company||Data processing|
|US8386485 *||Jul 31, 2009||Feb 26, 2013||George Mason Intellectual Properties, Inc.||Case-based framework for collaborative semantic search|
|US8402018 *||Feb 12, 2010||Mar 19, 2013||Korea Advanced Institute Of Science And Technology||Semantic search system using semantic ranking scheme|
|US8572161 *||Mar 12, 2008||Oct 29, 2013||Oracle International Corporation||Simplifying synchronization of copies of same data used by multiple applications|
|US8676787 *||Dec 22, 2011||Mar 18, 2014||International Business Machines Corporation||Distributed multi-step abstract queries|
|US8688683 *||Nov 30, 2009||Apr 1, 2014||Business Objects Software Ltd.||Query plan reformulation|
|US8725774 *||Oct 5, 2012||May 13, 2014||Xerox Corporation||Enforcing policies over linked XML resources|
|US8874552||Nov 28, 2010||Oct 28, 2014||Rinor Technologies Inc.||Automated generation of ontologies|
|US8918462 *||Jan 7, 2011||Dec 23, 2014||Cerner Innovation, Inc.||Using user-attribute ontologies to calculate user-anonymity degrees|
|US8943041||Sep 14, 2012||Jan 27, 2015||Business Objects Software Ltd.||Query plan reformulation|
|US8959082||Nov 30, 2011||Feb 17, 2015||Elwha Llc||Context-sensitive query enrichment|
|US9044543||Jul 17, 2012||Jun 2, 2015||Elwha Llc||Unmanned device utilization methods and systems|
|US9061102||Aug 31, 2012||Jun 23, 2015||Elwha Llc||Unmanned device interaction methods and systems|
|US9069816||Dec 12, 2012||Jun 30, 2015||International Business Machines Corporation||Distributed multi-step abstract queries|
|US9254363||Aug 31, 2012||Feb 9, 2016||Elwha Llc||Unmanned device interaction methods and systems|
|US9495345 *||Dec 9, 2015||Nov 15, 2016||Idibon, Inc.||Methods and systems for modeling complex taxonomies with natural language understanding|
|US9569439||Nov 30, 2011||Feb 14, 2017||Elwha Llc||Context-sensitive query enrichment|
|US20060224628 *||Jan 27, 2006||Oct 5, 2006||Bea Systems, Inc.||Modeling for data services|
|US20060271351 *||May 31, 2005||Nov 30, 2006||Danilo Mirkovic||Dialogue management using scripts|
|US20070174106 *||Jan 26, 2006||Jul 26, 2007||Chris Aniszczyk||Method for reducing implementation time for policy based systems management tools|
|US20090235185 *||Mar 12, 2008||Sep 17, 2009||Oracle International Corporation||Simplifying synchronization of copies of same data used by multiple applications|
|US20100057702 *||Aug 29, 2008||Mar 4, 2010||Oracle International Corporation||System and Method for Searching Enterprise Application Data|
|US20100070496 *||Sep 15, 2008||Mar 18, 2010||Oracle International Corporation||Searchable Object Network|
|US20100070500 *||Oct 31, 2007||Mar 18, 2010||Zhan Cui||Data processing|
|US20100070517 *||Sep 17, 2008||Mar 18, 2010||Oracle International Corporation||System and Method for Semantic Search in an Enterprise Application|
|US20100185643 *||Jan 20, 2009||Jul 22, 2010||Oracle International Corporation||Techniques for automated generation of queries for querying ontologies|
|US20100228782 *||Feb 26, 2009||Sep 9, 2010||Oracle International Corporation||Techniques for automated generation of ontologies for enterprise applications|
|US20100287158 *||Jul 14, 2010||Nov 11, 2010||Kinor Technologies Inc.||Information access using ontologies|
|US20110029514 *||Jul 31, 2009||Feb 3, 2011||Larry Kerschberg||Case-Based Framework For Collaborative Semantic Search|
|US20110131199 *||Nov 30, 2009||Jun 2, 2011||Business Objects Software Ltd.||Query plan reformulation|
|US20110173261 *||Jan 7, 2011||Jul 14, 2011||Cerner Innovation, Inc.||Using user-attribute ontologies to calculate user-anonymity degrees|
|US20110184972 *||Dec 23, 2010||Jul 28, 2011||Cbs Interactive Inc.||System and method for navigating a product catalog|
|US20110202526 *||Feb 12, 2010||Aug 18, 2011||Korea Advanced Institute Of Science And Technology||Semantic search system using semantic ranking scheme|
|US20120239677 *||Mar 15, 2011||Sep 20, 2012||Moxy Studios Pty Ltd.||Collaborative knowledge management|
|US20150100873 *||Oct 3, 2014||Apr 9, 2015||Orions Digital Systems, Inc.||Tagonomy - a system and method of semantic web tagging|
|US20150199389 *||Mar 24, 2015||Jul 16, 2015||Comprehend Systems, Inc.||Systems and methods for generating schemas that represent multiple data sources|
|US20160162476 *||Dec 9, 2015||Jun 9, 2016||Idibon, Inc.||Methods and systems for modeling complex taxonomies with natural language understanding|
|U.S. Classification||707/804, 707/807|
|International Classification||G06F, G06F17/30|
|Cooperative Classification||G06F21/6218, H04L63/0272, H04L63/10, G06F17/30463, G06F17/30569|
|European Classification||G06F21/62B, H04L63/10, G06F17/30S5V, H04L63/02C, G06F17/30S4P3T5|
|Jan 20, 2006||AS||Assignment|
Owner name: KINOR TECHNOLOGIES INC., ISRAEL
Free format text: ASSIGNMENT OF ASSIGNORS INTEREST;ASSIGNORS:TOLEDANO, ZEV;JEHUDA, JAIR;REEL/FRAME:017484/0525;SIGNING DATES FROM 20060117 TO 20060119
Owner name: KINOR TECHNOLOGIES INC., ISRAEL
Free format text: ASSIGNMENT OF ASSIGNORS INTEREST;ASSIGNORS:TOLEDANO, ZEV;JEHUDA, JAIR;SIGNING DATES FROM 20060117 TO 20060119;REEL/FRAME:017484/0525
|Jan 24, 2014||FPAY||Fee payment|
Year of fee payment: 4
|Jan 24, 2014||SULP||Surcharge for late payment|